Thanks @BFalquet , please see issues below,

1. I find this template incredibly hard to read and follow - is there no way to mimic the actual template structure proposed from GDS?

Many thnaks @BFalquet , please see issues below,

1. Treatment group needs (N=xx)
2. Any way we can get parameter at the top as by-group, rather than in the table?
3. Change from baseline should not be presented for Baseline
4. Statistics for n=0 should be presented as NE not NA
5. Column should be Value at Visit and not Analysis Value

and write examples that use it explicitly with tidyr functions

• Header should include in notes that BWGHTSI needs to be merged into ADSL before running
• Should the user be able to format the appearance of the variable label in Chevron (eg. Age to appear as "Age (yr)" and should we have the defaults built in?
• Should formatting of values and order of values be done inside or outside of Chevron (eg. Gender to appear as Male / Female) and should we have defaults built in?
• Template suggests a space between each variable presented, is this possible?
• If a variable has 0 count, it is dropped. Should there be some warning or message to user that this is happening?

Thanks @BFalquet , please see feedback below

1. There is no need to call in ADAE - this should be removed
2. When running with ADAE I get following error? Error: Assigned data varvec must be compatible with existing data. Existing data has 18 rows.Assigned data has 0 rows.Only vectors of size 1 are recycled.

• aet01_1
• aet02_1
• aet03_1

see other NEST projects for an example

do not save the docs folder, i.e. add it to the .gitignore

• No titles appearing (DMT01 has titles, should we expect appear in all?)
• Can we have space between each SOC?
• Should header first describe the 3 versions available at top?
• If data not available a null report should be generated (and not error)

High Priority

1. How will formatting of Trt Groups be applied (incl. wrapping suggestion if needed)? Outside of Chevron?
2. I still like the idea of defining in Chevron for each template whether we expect Planned or Actual Trt, and this be defined with a var at a global level - could we discuss?
3. Each dataset needs df_explicit_na applied to them - need a dm_explicit_na - and where is this done?
4. Set-up of dm and keys could be standardised, part of Chevron? Also a new way for many to be working with data manipulation - this ok?
5. COVID filter flag is based on ADSUB - do we need to think how users would attach to ADSL ahead of filtering?
6. Reporting conventions not being followed - do we strictly apply or discuss/challenge? Example display of NE when stat not computable, dp one more for mean/median/SD, etc.
7. General - where is filtering / titles+foots being done now, in Chevron or outside and how much do we need to be testing?

Low Priority

1. If a number is to 1d.p. (eg. %), can we set to 25.0 rather than limiting down to 25? Is this an R or rtables limitation?
2. Can we get alignment within brackets (eg. show 2.5% in brackets as ( 2.5%) rather than (2.5%))
3. If filtered data does not generate a report, a null report is required to be produced - is that Chevron or outside?
4. If data is not present in a treatment group, how can we still control presentation of the treatment group (eg. AE tables to show 0 events)?
5. Flexibility of label of Patient to be changeable to Subject or Participant?
6. Is there a background dataset that can be put forwards for QC?

@insightsengineering/chevron

• basic_table_deco
• lbl_overall should be NULL by default

include dst01, dmt01 and aet02

When binding 2 tables, the title and decorator disappear.

Goal

To implement AET01 where

• the analysis population: AP / IT / Any population filter - Default SE
• Column Variables: TRT01P / ARM / TRT01A / Any ADSL variable - Default ACTARM
• Column Totals: All Patients / Any pooling - Default NONE
• Analysis Concepts: "Severe AE" interchangeable with "Grade 3-5 AE"
  Incorporate any variable/flag from ADAE

• Default list
  Total number of patients with at least one AE
Total number of AEs
Total number of deaths
Total number of patients withdrawn from study due to an AE
Total number of patients with at least one
AE with fatal outcome
Serious AE
Serious AE leading to withdrawal from treatment
Serious AE leading to dose modification/interruption
Related Serious AE
AE leading to withdrawal from treatment
AE leading to dose modification/ interruption
Related AE
Related AE leading to withdrawal from treatment
Related AE leading to dose modification/interruption
Severe AE (at greatest intensity)

Note

Above to be default AET01_1 with preprocessing step adding these temp flags to adsl and user selecting what they want to keep. Flexibility for user to add own categories if study specific variations
AET01_2 includes Medical Concepts section which would be user-driven and based on SMQ/CQ adae variables

write a vignette on why layouts are useful -- as currently its not clear from the documentation

... from the tlg-catalog

EXT01

• Example shown in vignette is not accurate, also ANL01FL is not relevant subset
  (see n, maybe filter by drug in PARCAT2, one table by drug)
  (filter for PARCAT1 == "OVERALL" )
• How are we defining the variables to include? Is this by user filtering?
  => filtered inside the function (for now)
• I do not understand EXT01_2, think we need a rethink on how AVALCAT is included?

LBT01

• Missing N in treatment header
  => should be under ACTARM (not CHG or AVAL)
• How is precision controlled/will be controlled? Different tests require different precision
  => add info to study object ?
  "A reasonable level of precision for means, SDs, and medians is one digit more than the level of precision of the test. Present the minimum and maximum to the level of precision of the test."
• Column headers do not match standard

... as argument if the dataset is only used in alt_count in build_table

to be started and discussed

I suggest to start with listing the different study objects

• efficacy
• safety
• etc...

and then list all the study object members that we currently have and their meaning.

the header disappears when tables are bound together.
Use the logic:

t2 <- basic_table(title = "my title", subtitles = "my subtitle", main_footer = "main footer") %>% split_cols_by("ACTARM") %>% add_colcounts() %>% add_overall_col(label = "All patients")
t2 %>% add_existing_table(t1) %>% build_table(adsl)

the problem exists in

• aet01_1
• dst01_1
• dst01_2
• dst01_3

to have adsl as first argument in every table generating function (when necessary).

• in bash navigate to R folder of package
  for x in *.R ; do sed -i 's/|>/%>%/g' "$x" ; done

• import magrittr pipe operator

its feature that is not available in rtables right now but on the roadmap.

Thanks @BFalquet , please see issues below,

1. Example looks wrong, n too high?
2. When running on my study data i get following error - Error in assert_colnames(adex, summaryvars) : Variable(s) not a column name in adex : AVAL

Title & subtitles generated by the decorator function are still WIP.
The current system to exclude total column inadvertently removes the title section.

.. see tern

ask Nik Burkoff

• update DESCRIPTION
• update README.md
• use |> instead of %>%
• update R CMD check automation

format (type) description or (type)\cr description

improve documentation:

• dst01_x
• dmt01_x
• aet02
• aet03

R CMD Check needs to 0 Errors 0 Warnings 0 Notes

so that automation passes

Modify dmt01 to provide an understandable error message when the variable names passed to the summaryvars argument of are not present in the dataset passed through the ad_bl argument.

Should the program default to existing variable names or halt execution?
=> halting execution is probably safer.

i.e. update the @details and discuss with Tim

for the tables dst01_1, dst01_2, and dst01_3 make meaningful titles and description:

• dst01_1
• dst01_2
• dst01_3

use the same markdown documentation